There is evidence from the published evaluation reports that the National Youth Anti-Drug Media Campaign (NYAMC) produced boomerang effects of exposure to anti-drug ads on cognitive and-behavioral marijuana-related outcomes. More exposure to the ads led to pro-drug beliefs. The proposed research will investigate potential explanations for the apparent effects. Proposed mediators-and moderators are based in health behavior theory. The findings of such research will enable achievement of the long-term study goals: to understand factors that bring about boomerang effects, so as to inform the design of public health interventions including mass media campaigns, particularly future anti-drug efforts. Two theories (Social Norms theory and the Theory of Reasoned Action)-offer mediating mechanisms for adverse effects and the remaining three theories (Elaboration Likelihood, Psychological Reactance, and social network theories) suggest message and individual response differences that may have moderated the boomerang effects. We hypothesize that the effects of exposure on marijuana intentions were mediated by 1) social norms, specifically youth perceptions of peer marijuana use prevalence, and 2) outcome expectancies, specifically youth beliefs about positive consequences of marijuana use. Thus, we predict exposure to NYAMC ads produced stronger beliefs in widespread marijuana use among their peers and; in favorable outcomes of marijuana use which both, in turn, led to greater intentions to use marijuana. We also hypothesize that the boomerang effects were moderated by audience evaluations of 3) the anti-drug ads' argument strength and 4) exaggeration of the problem of youth marijuana use, and by 5) the frequency and the valence of the conversations youth have about drugs with others including peers and siblings. We propose to test these possible explanations using the NSPY data. We will examine the mediating hypotheses by examining the extent to which the exposure-outcome relationship is reduced when the mediating variables are statistically controlled; we will examine: the moderating hypotheses by examining whether exposure and the moderators interact in their joint effects on marijuana related outcomes; We will focus on delayed effects to avoid ambiguity of causal direction and control for possible confounders using counterfactual weights as well as more commonly used regression procedures. Almost all analyses will use Wesvarp to generate correct population estimates. Explanations for adverse effects of public communication campaigns are presently understudied. However, without such research, progress can not be made in terms of improving the effectiveness of this public health approach and avoiding such pitfalls in the future.